I went to the doctors today and before I got a chance to speak with the doctor she handed me a piece of paper with information about something called Heidi and asking for my consent for its use.

It turns out Heidi is a product from a company called Lyrebird health

No, I’d never heard of them either but they seem to provide health professionals with transcription services. Their governance seems to aligned with Australian and European requirements but I couldn’t find any information about what transcription model they use and how it was trained. Only that they don’t train on my data and that all inference happens locally in our region (which suggests it’s running on AWS). If they don’t train on our data, I wonder then what data it is trained on?

I was surprised at how I felt through all this. On one hand I gave my consent because I wanted to see what happened but on the other, I was very aware that there was a third, invisible party in the room listening, recording and considering my conversation. In a world where it’s now totally conceivable that all data is accessible by bad actors, the jury is still out for me on this.

Following on from my previous post about how AI is starting to fall out of favour with people, I can see how interactions like the one I had today are contributing to a growing sense of unease. At what point does the doctor become unnecessary and given appointments are already harder to find than hens teeth are doctors sleepwalking into solving shortages with their obsolescence at least for many of the lighter touch issues that they’re already pushing people to telephone consults for post covid.

Next visit I’ll have to try retracting my consent and see what happens.